Results 1 -
4 of
4
Active Learning in Social Context for Image Classification
"... multimodal fusion Abstract: Motivated by the widespread adoption of social networks and the abundant availability of user-generated mul-timedia content, our purpose in this work is to investigate how the known principles of active learning for image classification fit in this newly developed context ..."
Abstract
-
Cited by 2 (2 self)
- Add to MetaCart
(Show Context)
multimodal fusion Abstract: Motivated by the widespread adoption of social networks and the abundant availability of user-generated mul-timedia content, our purpose in this work is to investigate how the known principles of active learning for image classification fit in this newly developed context. The process of active learning can be fully automated in this social context by replacing the human oracle with the user tagged images obtained from social net-works. However, the noisy nature of user-contributed tags adds further complexity to the problem of sample selection since, apart from their informativeness, our confidence about their actual content should be also max-imized. The contribution of this work is on proposing a probabilistic approach for jointly maximizing the two aforementioned quantities with a view to automate the process of active learning. Experimental results show the superiority of the proposed method against various baselines and verify the assumption that significant performance improvement cannot be achieved unless we jointly consider the samples ’ informativeness and the oracle’s confidence. 1
Using tagged images of low visual ambiguity to boost the learning efficiency of object detectors
- In ACM Multimedia
, 2013
"... Motivated by the abundant availability of user-generated multimedia content, a data augmentation approach that en-hances an initial manually labelled training set with regions from user tagged images is presented. Initially, object detec-tion classifiers are trained using a small number of manually ..."
Abstract
-
Cited by 1 (0 self)
- Add to MetaCart
(Show Context)
Motivated by the abundant availability of user-generated multimedia content, a data augmentation approach that en-hances an initial manually labelled training set with regions from user tagged images is presented. Initially, object detec-tion classifiers are trained using a small number of manually labelled regions as the training set. Then, a set of posi-tive regions is automatically selected from a large number of loosely tagged images, pre-segmented by an automatic seg-mentation algorithm, to enhance the initial training set. In order to overcome the noisy nature of user tagged images and the lack of information about the pixel level annota-tions, the main contribution of this work is the introduction of the visual ambiguity term. Visual ambiguity is caused by the visual similarity of semantically dissimilar concepts with respect to the employed visual representation and analysis system (i.e. segmentation, feature space, classifier) and, in this work, is modelled so that the images where ambiguous concepts co-exist are penalized. Preliminary experimental results show that the employment of visual ambiguity guides the selection process away from the ambiguous images and, as a result, allows for better separation between the targeted true positive and the undesired negative regions.
HOWMANYMORE IMAGES DOWE NEED? PERFORMANCE PREDICTION OF BOOTSTRAPPING FOR IMAGE CLASSIFICATION
"... Motivated by the recently introduced scalable concept de-tection challenge that requires classifiers for hundreds or even thousands of concepts, the objective of this work is to predict the cases where the enhancement of an initial classifier with additional training images is not expected to provid ..."
Abstract
- Add to MetaCart
(Show Context)
Motivated by the recently introduced scalable concept de-tection challenge that requires classifiers for hundreds or even thousands of concepts, the objective of this work is to predict the cases where the enhancement of an initial classifier with additional training images is not expected to provide signif-icant improvements. To facilitate this objective, we need a model for predicting the performance gain of a bootstrapping process prior to actually applying it. In order to train this model, we propose two features; the initial classifier’s matu-rity (i.e. how close is the current hyperplane to the optimal) and the oracle’s reliability (i.e. how reliable is the oracle in providing the correct labels of new training data). Thus, the contribution of our work is on proposing a method that is able to exploit the correlation between the expected performance boost and these two indicators. As a result, we can consider-ably improve the scalability properties of such bootstrapping processes by concentrating on the most prominent models and thus reducing the overall processing load. Index Terms — scalable concept detection, image classi-fication, performance prediction, bootstrapping 1.
